cancel
Showing results for 
Search instead for 
Did you mean: 

Sybase replication server on HA / DR

Former Member
0 Kudos

Hi Experts,

We are in the process of setting up DR environment for Sybase ASE using replication server. The primary site has an HA configured.

My doubt is when we need to install replication server at standby site, what host needs to be specified as the "primary system hostname" - The hostname of CI or the DB server.

Additionally, do we need to install replication server on both nodes of HA cluster? and how would the DR agent configuration work in case of DB failover to the CI node?

Unfortunately, there is very little documentation available online on how to setup the DR environment, apart from SAP note 1891560

Would appreciate your invaluable inputs on this.

Thanks in anticipation.

Regards,

Varun

Accepted Solutions (1)

Accepted Solutions (1)

Former Member
0 Kudos

Hi Varun,


Can you share with us what kind of HA solution is being used?

In general:

You always want to use the name of the 'cluster' for any host name (assuming this name or IP will always point to the currently active node in the HA environment.

For installation, without knowing what your HA solution looks like, I would say "follow the ASE".

Install the replication software on the same hosts and same locations as the ASE software.  If you are installing ASE on a shared storage device between the HA nodes, you will want to do the same with the replication software.  If you are installing the Replication Server on a totally different host (not part of the HA solution), then this would not apply.

Regards,
Stephen

Former Member
0 Kudos

Hi Stephen,

We have AIX 7.1 OS and use powerHA (HACMP) as our HA solution. Node A has CI and ASCS while node B has Sybase Database and DI.

ASCS and Database are installed on shared storage device.

Since this setup points Sybase ASE DB to a single node at a time,  and we do not have a service hostname, I am not sure how we can "follow the ASE".

Any thoughts ?

Regards,

Varun

Former Member
0 Kudos

Hi Varun,

Sorry - I don't have any experience with the IBM solution.  So I have comments, but not a solution.

The HADR Solution will work well in HA / cluster systems where the IP address is dynamically re-assigned to the active node.  This allows the server configurations to point to just 'one' IP address and the cluster software (examples: MS Cluster, Veritas, HP Service Guard) assigns that IP to which ever node is active.

Using the IBM solution, (based on a quick Google search) it appears we rely on Open Server HA capability to include both addresses in the interfaces file under a single Server name, so that both address can be attempted when making a connection.

Unfortunately, the configurations created by the DR Agent do not support this type of configuration using Interfaces files.  Most server address are assigned a single IP address, and most address are assigned directly (not pointing to an Interfaces file entry).

As an example, when the HADR nodes are configured in ASE, you define the other nodes in the HADR group with a command like the following:

sp_hadr_admin addserver, <node_name>,<server_name>

To use a interfaces file name for the server name, the command might look like:

sp_hadr_admin addserver, 'D01_Site1', 'D01'

But the configurations created by the DR Agent use a host:port syntax instead of an interfaces file server name, like:

sp_hadr_admin addserver, 'D01_Site1', 'host1.my.domain.corp:5001'

The point I am trying to make is that even if we edit the interfaces files to use the additional HA addresses, other configurations like the one above would need to be discovered and changed as well.

Unfortunately, even if all this reconfiguration work is accomplished successfully, the DR Agent itself does not use interfaces files and has no support for specifying an alternate HA address.  And I am not aware of any workaround for that limitation.

In summary, I think that short answer is that we don't have HADR support using DR Agent for HA environments without a dynamic IP address.  It might be possible to manually reconfigure some of it, but without support in the DR Agent itself, you would be missing failover and monitoring support the DR Agent provides.

Sorry that I do not have a more encouraging answer.

Regards,
Stephen

Former Member
0 Kudos

Hi,

I Encountered the issue during installation of replication server on windows cluster node.

My environment is

Primary SAP HA on windows cluster and

Standby SAP Standalone on standby side.

i got the error

Assertion failed: Replication Server and Database server on same host is not supported in OS cluster environment

So my solution was install replication services alone with DR agent on saprate HOST, out side the cluster on primary side and install replication service on same host at standby side.

Working Solution is

Primary

SAP HA on windows cluster   

NEW HOST/SERVER with  Sybase replication server

Standby SAP Standalone on standby side + Sybase replication on same host.

And it work for me successfully, I asked SAP support to validate this solution but i didn't get any acknowledgement yet, But its working for me since 4 months.

Sumit

Answers (0)